Bizarre issues with nvlddmkm driver failure

[fishyfishyfishy]

A system I've been working on keeps getting the nvlddmkm.sys driver failure over and over while under moderate stress (ie in a game) and continues to fail and recover until it inevitably BSODs. I doubt it's a power issue as the system has a rock-solid corsair 750w PSU running 2 9800 GTs and some flavor of quad core (maybe Q6600, I can't remember). The ram is corsair xms2 and although I haven't tested it yet I've seen no evidence that makes me suspect that it's the problem. What's strange is that the machine will BSOD when both 9800s are in, whether or not they are in SLI. With one of the cards in the system seems stable, but when I run it with the other it starts the cycle of BSODing. Identical cards with identical drivers. What gives? Is one of the cards perhaps bad? If not, why would one screw up and not the other? Thanks for the help.

[TheTerrorist_75]

Are you overclocking the graphics cards? Do you know the temperatures of the CPU and graphics? Could be caused by wrong voltages or overheating. You can download and install HWMonitor to view them.

[fishyfishyfishy]

The cards aren't overclocked. CPU temp is below 60 C at all times (rarely gets to 50) and usage rarely maxes. The graphics temp doesn't exceed 76C (sits around 74 under moderate load). The temps and usages are the same regardless of the number of cards or configuration (which is a bit counter intuitive).

The system hadn't been cleaned before I looked at it and after cleaning it does run games a bit longer before BSODing, but only by a few extra seconds.

[TheTerrorist_75]

I would monitor the voltage to the graphics cards to see if they are staying stable.

[YoKenny]

I see that HWMonitor includes Ask.com Toolbar pre-selected for installation.

I prefer SpeedFan